Anthropic's Q3 Revenue Gain Fell to $5.2 Billion From $6.9 Billion in Q2
A.I. / news
Anthropic's Q3 Revenue Gain Fell to $5.2 Billion From $6.9 Billion in Q2
Figures shown to prospective investors on Oct. 14 put the quarter at $16.8 billion, with an 11% adjusted operating margin Anthropic has not defined.

Anthropic told prospective investors on Oct. 14, 2026 that third-quarter revenue reached $16.8 billion, up about 45% from $11.6 billion in the second quarter. The dollar gain shrank to $5.2 billion from about $6.9 billion a quarter earlier, according to Implicator.ai's account of the meeting.
The figures were shared privately and are not audited. Anthropic declined to comment, Implicator.ai reported.
The quarterly figures
Implicator.ai puts first-quarter revenue at $4.7 billion. Henry Blodget, writing on his Substack Regenerator, cites a New York Times report and gives $4.8 billion for the same quarter. He calculates second-quarter growth at 141%. Implicator.ai says about 147%. Both give $11.6 billion and $16.8 billion for the next two quarters, so the disagreement is confined to the first quarter and the growth rate that depends on it.
- Q14.7 $ billion
- Q211.6 $ billion
- Q316.8 $ billion
Source: Implicator.ai, Oct. 15, 2026, citing figures shown to investors on Oct. 14; accessed 2026-11-03
Blodget's post argues the slowdown happened in August and September. He converts a $65 billion annualized run rate reported for July into about $5.4 billion for the month. That leaves about $11.4 billion for the final two months, or $5.7 billion a month. This is his arithmetic on reported numbers, and he warns that treating a run rate as one month's revenue may not hold.
Margins, compute and the IPO
Implicator.ai reports an adjusted operating margin of 11%, the second straight quarter of adjusted profit, and revenue of $1.70 for every dollar spent on computing power. Anthropic has not said which costs the adjusted figure leaves out.
On a standard accounting basis, the 2025 net loss was $42 billion. About $34 billion of that came from charges tied to past fundraising, per the same account. The operating loss excluding those charges exceeded $8 billion.
SemiAnalysis estimates annualized revenue at about $74 billion, up from about $65 billion in late July, according to Implicator.ai. Analysts who met executives put the year-end run rate near $110 billion. Some backers had hoped for more than $100 billion.
What the slowdown could mean
Blodget offers three explanations and establishes none. August is seasonally slow. Anthropic may have been unable to supply enough capacity. Or demand is moderating as cheaper Chinese models take share, a claim he makes without data. Blodget also notes Anthropic gave no September run-rate at the meeting, per DealBook.
An early backer told Fortune that "Anthropic can afford to slow down and still have huge value," per Implicator.ai. That backer holds a stake, which colours the view.
Revenue figures say nothing about the models. For those, see the site's earlier pieces on an architect's permit-drawing test of Claude Opus 5.5 and on Anthropic's five exploit chains.
The IPO timetable
Anthropic submitted a draft registration statement to the SEC on June 1, still unpublished. Its last private raise was $65 billion at a $965 billion post-money valuation in May. Some prospective investors put fair value at $1.8 trillion to $2 trillion.
Implicator.ai says marketing could begin the week of Nov. 9, with trading targeted for Nov. 26, after the Nov. 3 midterm elections. It calls the timing fluid.
Sources
More in A.I.
- 01Qwen3.8-27B Has 8 Million Downloads and Only Alibaba's Own BenchmarksThe Apache 2.0 model's SWE-bench Pro lead over Claude Opus 4.6 Max rests on a benchmark Alibaba corrected itself, and no outside party has scored it.
- 02GPT-Rosalind-6 Reuses GPT-6 Astra's Weights, Cuts Biology Refusals to 9.4% and Stays Behind a GateOpenAI's Oct. 30 system card addendum says no training was added, so the difference from Astra is deployment controls and who is allowed in.
- 03Architect Says Claude Opus 5.5 Drew a House Set Good Enough for a Permit After Two Redline RoundsStephen Gallagher's informal test of 13 models gave Opus 5.5 the most human feedback, and the drawings still missed a required second exit.
- 0450 GPT Agents in Minecraft Built a Ration Ledger, and 34 Died Doing ItY Combinator's Labs group and Olam Labs ran GPT-6.1 Sol and Claude Opus 5.5 swarms through Minecraft and Mindustry. GPT won both head-to-head games, but each result comes from one harness and one setup.